無線感測網路下基於所在房間位置之高效率室內定位技術

Abstract

隨著無線網路及超大型積體電路技術的快速發展，有更多的無線感測網路應用，如利用無線感測網路來提供病患的位置資訊，以協助醫療照護就是很好的例子。在室內的環境中，提供所在房間位置資訊比提供位置座標資訊更有參考價值。雖然目前基於訊號強度之定位演算法可達到不錯的準確度，但其利用座標資訊直接轉換為房間資訊卻常有一定程度的誤差。在本論文中，我們提出在無線感測網路下基於所在房間位置之高效率室內定位技術，其主要目標為提供高準確率之房間定位資訊以及降低演算法的計算成本。本研究藉由地圖資訊，限制標地物可移動的位置，並結合以座標資訊為基礎之特徵值比對演算法來減少無線網路多重路徑訊號影響所造成房間位置資訊的誤判，並降低定位誤差。實驗結果顯示，在RADAR定位系統上應用我們的方法可以將房間資訊的定位準確度自69% 提升到94%，並且將定位誤差由3.19公尺降低為2.24公尺。我們也與以提供房間位置資訊的定位演算法Signpost做比較，實驗結果顯示雖然我們的方法會比Signpost多了一秒的定位延遲時間，但在各種的干擾環境下，我們的房間資訊的定位準確度比Signpost增加20%。

With rapid advances in wireless and VLSI technologies, applications of wireless sensor networks (WSN), such as target localization in medical environments, are getting more and more popular. In an indoor environment, identifying a target in a logical area, such as a room is more meaningful than identifying a location represented by coordinates. In existing indoor localization algorithms, although their average location estimate errors may be low, they may not provide high accuracy room-level information, which is usually transformed from the predicted coordinates obtained by their localization algorithms. In this thesis, we present a room-based indoor localization scheme using a path-restricted room-level localization algorithm (PRLA) and a fingerprinting based technique for wireless sensor networks. The design of the proposed PRLA localization system meets two objectives. The first is high accuracy of room-level localization and the other is low computational overhead. Experimental results show that by applying our PRLA scheme to the RADAR system, it significantly improves its room-level hit rate from 69% to 94% and reduces the average error distance from 3.19 m to 2.24 m. In addition, the proposed PRLA scheme can be applied to any coordinate-based indoor location system to enhance its room-level hit rate and localization accuracy. We also compared the proposed PRLA with Signpost which is also a room-based localization scheme. Experimental results demonstrate that the delay time to enter a room of the proposed PRLA was one second more than that of Signpost; however the proposed PRLA is more reliable than Signpost because its average room-level hit rate is 20% better than Signpost’s in our experiment when we move around in the same room.